Jovian UV Aurora's Response to the Solar Wind: Hisaki EXCEED and Juno Observations
Abstract
We summarize Jupiter's ultraviolet (UV) auroral response to solar wind dynamic pressure variations during Juno's approach to Jupiter in 2016. The response time of Jupiter's aurora to external drivers has thus far been unknown owing to a sparsity of upstream in situ solar wind measurements. Combining the Juno solar wind observations with continuous UV aurora data obtained by Hisaki EXCEED (Extreme Ultraviolet Spectroscope for Exospheric Dynamics) and Juno UV spectrograph, the UV aurora brightenings in response to three major shock arrivals showed time lags of 10–15 hr. These time lags are longer than the time required for ballistic propagation of the shocks by the solar wind. In addition to that puzzle, while an enhancement in the UV auroral power was observed with an increase in dynamic pressure to ~0.03 nPa, no associated brightening was observed with a dynamic pressure elevation of >0.1 nPa. These imply that internal magnetospheric aspects need to be taken into consideration to fully resolve the issue.
Key Points
- We compare Jupiter's ultraviolet aurora variation observed by Hisaki with changes in the upstream solar wind conditions observed by Juno
- Transient brightenings responded to major solar wind shocks with ~10 hr lag time, which is inconsistent with a solar wind propagation model
- A brightening triggered by a dynamic pressure elevation of 0.03 nPa was detected, whereas a 0.1 nPa elevation did not trigger a brightening
Plain Language Summary
Jovian ultraviolet aurora are emitted from hydrogen molecules in Jupiter's atmosphere when energetic electrons precipitate from the magnetosphere to excite the atmospheric molecules. The Jovian magnetosphere is always under the influence of the solar wind. Variation in the solar wind affects magnetospheric dynamics and thus the Jovian aurora intensity. The solar wind-magnetosphere interaction is well studied for Earth, and the issue of aurora response to the solar wind is also well studied for Earth, but the issue remains open for Jupiter. Here we obtain the response time of aurora brightening upon intensification of the solar wind, which is a very fundamental quantity, to find it to be too long to be explained by a simple propagating model that assumes the solar wind as the dominant driver. Furthermore, some small variations in solar wind shocks led to aurora brightenings, while larger variations did not trigger other events. The characteristics discussed in this paper provide good case studies to validate theories or numerical simulations of how Jovian aurora may respond to changes in the solar wind.
1 Introduction
Jupiter's magnetosphere-ionosphere (M-I) coupling current system is generated by the combination of the plasma supplied from Jupiter's moon Io and the fast rotational motion of the planet (Hill, 1979). In addition, there is solar wind compression of the magnetosphere on the M-I current system (Chané et al., 2017; Cowley et al., 2007; Cowley & Bunce, 2001, 2003a, 2003b; Southwood & Kivelson, 2001). Cowley et al. (2007) modeled auroral currents near the open-closed field line boundary finding that they increase during the rapid compression phase, which lasts approximately 2–3 hr, and then decrease in the steadily compressed magnetosphere, which appears 1–2 days after the start of the event. Chané et al. (2017) investigated the solar wind influence on the aurora region by three-dimensional global magnetohydro dynamics (MHD) simulations. They found that when the high-density solar wind arrives at the magnetopause, the auroral currents start changing after ~2.5 hr and reach a new equilibrium after 60 hr.
According to previous studies, the solar wind influences Jupiter's auroral radio (Hess et al., 2012, 2014; Nakagawa et al., 2000; Terasawa et al., 1978), infrared (Baron et al., 1996), ultraviolet (UV; Badman et al., 2016; Prangé et al., 1993; Waite et al., 2001; Kita et al., 2016), and X-ray (Dunn et al., 2016; Kimura et al., 2016) emissions. Solar wind compressions that cause significant disturbances in the terrestrial magnetosphere also cause aurora storms in the Jovian and/or Saturnian magnetospheres (Oya & Morioka, 1981; Prange et al., 2004). The solar wind conditions at Jupiter are typically extrapolated from near-Earth measurement (e.g., Tao et al., 2005), since in situ solar wind measurements at Jupiter are limited (e.g., Ebert et al., 2014; Jackman & Arridge, 2011). Spacecraft that have observed the solar wind near Jupiter include the Pioneer, Voyagers, Ulysses, Cassini, New Horizons, and Juno. Solar wind measurements from Voyager 1 and 2 correlated HOM (HectOMetric) (Desch & Barrow, 1984) and Non-Io DAM (DecAMetric) radio emissions (Barrow et al., 1986). During Ulysses first flyby in February 1992, the infrared H3+ emission intensity from both poles had a positive correlation with the solar wind dynamic pressure (Baron et al., 1996; Connerney et al., 1996). Ulysses re-encountered Jupiter in February 2004 at a distance of 1,684 Jovian radii (RJ). The non-Io DAM emission obtained from Nançay correlated with solar wind dynamic pressure enhancements propagated from Ulysses to Jupiter during that timeframe (Echer et al., 2010). Cassini observed in situ solar wind conditions near Jupiter in January 2000. The intensity of HOM radio (Gurnett et al., 2002; Prangé et al., 2004) and ultraviolet (UV) auroral (Pryor et al., 2005) emissions increased around the arrival time of a solar wind shock. Seven Hubble Space Telescope (HST) observations were made during the Cassini flyby. During one observation, a brightening of the main oval corresponded with the arrival of a solar wind compression region (Nichols et al., 2007). An HST campaign was also conducted during the New Horizons Jupiter flyby (Clarke et al., 2009; Nichols et al., 2009). They used the propagated solar wind data from Earth measurements and shifted the time axis by 2.1 days in order to fit the arrival time of an interplanetary shock observed by New Horizons. The total auroral power increased around the arrival times of these shocks.
The extrapolated solar wind models described above are useful; however, they have a large uncertainty in estimating a shock arrival time (e.g., about ±20 hr if the Earth-Sun-Jupiter angle is less than 50°, Tao et al., 2005). While the solar wind data obtained around Jupiter would highly reduce the arrival time ambiguity, such solar wind plasma data have been limited and, at times, sparse. As can be seen in Nichols et al. (2007), the Cassini solar wind plasma data had 3- to 4-hr gap. Such ambiguity, sparsity, and noncontinuity of the solar wind information prevented us from conducting a detailed, long-term analysis of Jupiter's auroral response to solar wind shocks. For example, the time lag between the solar wind shock arrival and aurora onset has never been reported. The relationship between the size of dynamic pressure enhancement and the aurora variation is also unknown. These parameters reflect the magnetospheric response to the solar wind. Since the aurora response to the solar wind is not simple, such parameters are crucial to test theories and numerical simulations. In 2016, Juno approached Jupiter from the dawn side and continuously observed the solar wind plasma for ~40 days (McComas et al., 2017; Wilson et al., 2018). At the same time, the Hisaki satellite, as well as Juno, continuously monitored Jupiter's UV aurora (e.g., Gladstone et al., 2017).
Here we compare the time evolution of the Jovian UV aurora obtained by Hisaki/Juno and the solar wind dynamic pressure observed by Juno. These coordinated observations provide an unprecedented opportunity to examine how Jupiter's UV aurora responds to variable solar wind conditions with little ambiguity as to the external driver. The detailed time evolution of the aurora during the solar wind compression contains significant information on how Jovian magnetosphere responses to the change in the solar wind conditions. Mainly we focus on the shock arrival time and the size of dynamic pressure enhancement associated with the aurora enhancement.
2 Instruments and Data Analysis
Hisaki is an Earth-orbiting spacecraft equipped with a UV spectroscope that primarily observes planetary atmospheres and magnetospheres. The Hisaki EXCEED (EXtreme ultraviolet spetrosCope for ExosphEric Dynamics) (Yamazaki et al., 2014; Yoshikawa et al., 2014; Yoshioka et al., 2013) spectrometer has a unique dumbbell-shaped slit for Jupiter. It is designed to observe the Io plasma torus and Jovian aurora simultaneously. The spatial resolution is 17 arc sec (Yamazaki et al., 2014; Yoshikawa et al., 2014); therefore, we cannot spatially resolve the Jovian aurora. In order to obtain the time series of rotationally averaged power of Jovian UV aurora, we follow the same procedure as Kita et al. (2016). The EXCEED level-2 spectrum (Kimura et al., 2019) was averaged over 10 min and 900–1,480 Å where geocoronal emission lines were excluded. We also excluded the data when the central meridian longitude was smaller than 100° and larger than 250° because the northern aurora is difficult to see from Earth. The central meridian longitude dependence of the total auroral power was corrected by fitting a sinusoidal function (Kita et al., 2016). We finally averaged the total auroral power over one Jovian rotation to construct a longitudinally averaged total power.
The Jovian upstream solar wind plasma parameters were provided from the Jovian Auroral Distributions Experiment (McComas et al., 2017) ion sensor (JADE-I) on board the Juno spacecraft (Wilson et al., 2018). JADE solar wind data is available from day of year (DOY) 136 to 176 in 2016 when Juno was in interplanetary space near Jupiter. The solar wind velocity, density, and ram pressure are available at a cadence between 30 and 600 s. The shock arrival time at Juno and Jupiter should have a lag time because the heliospheric position is slightly different. The difference of heliospheric radial distance between Juno and Jupiter varied from 2.6 × 106 to 3.7 × 105 km, and the Juno-Sun-Jupiter angle varied from 2.2° to 0.66° over the period. We used a one-dimensional MHD model (Tao et al., 2005) to estimate the difference in solar wind arrival time between Juno and the center of Jupiter (note that this estimation does not consider the effects of the magnetosheath and the magnetosphere). Because Juno was located downstream of Jupiter with respect to the solar rotation, any corotating interaction region (CIR) would hit Jupiter first. CIR could be identified by the time profile of the solar wind speed and density. A heliospheric current sheet crossing was also observed near CIR. The estimated lag time for CIR varied from 2.1 to 0.72 hr. In the case of interplanetary coronal mass ejection (ICME), the lag time mentioned above should be different because a shock front of ICME is not aligned along with a typical spiral structure of the solar wind. ICME could be identified by the rotation in the interplanetary magnetic field angles and the large variance in the proton beta. If the solar wind shock was associated with an ICME, the arrival time at the center of Jupiter was estimated using the solar wind velocity and the radial difference between Sun-Juno and Sun-Jupiter.
Hisaki UV auroral total power and the solar wind dynamic pressure at Juno are shown in Figures 1a and 1c, which is averaged over one Jovian rotation (~10 hr) and 1 hr, respectively. Juno observed three significant structures around DOY 141 (ICME), DOY 150 (CIR), and DOY 173 (CIR) (McComas, Szalay, et al., 2017; Nichols et al., 2017). Because of the degradation of the Hisaki/EXCEED field of view, it is more difficult to track the target with the guide camera after DOY 170. Because of the aurora emission as well as the contamination by the disk emission, the total power does not fall below ~250 GW if the tracking works correctly. The total auroral power decreased below ~250 GW after DOY 170, which turned out that the northern aurora moved outside the slit. Therefore, one did not use Hisaki aurora data after DOY ~170. For this reason, Juno/UVS data (Gladstone et al., 2017) is used to discuss the variation in the total auroral power during the third solar wind shock.
The approach phase observations of Juno/UVS are summarized in Gladstone, Versteeg, et al. (2017). Because of the limited spatial resolution during the approach, Juno/UVS only provide a spatially integrated total auroral power for the north and south pole. Integrated wavelength ranges are 700–1,190 Å and 1,230–1,620 Å, where the Lyman α emission line is excluded because of the limited data volume. The total emitted auroral power is estimated by the above-integrated value multiplied by 1.2 to correct for the missing fraction of the H2 band and Lyman α emission (Gladstone, Versteeg, et al., 2017). The projected lengths of the aurora ovals defined by the VIP4 model are used to correct the 10 hr of rotational modulation in the auroral power. Residuals from 10-hr modulation are ~20% after applying this process; however, it is still useful to reduce the modulation.
Figure 1b shows the Juno UVS total power of the northern (circles) and southern (crosses) aurora from UVS that are integrated over one Jovian rotation, in the same format as the Hisaki data. The total power from Hisaki data is superposed with the dashed line. The total power from Hisaki data is superposed with the dashed line. The total emitted power obtained from Juno/UVS are ~2.5 time smaller than Gladstone, Versteeg, et al. (2017), mainly due to calibration revisions in the effective area. During the quiet period, the total powers from Hisaki EXCEED and Juno UVS are as large as 250 GW and 2 TW, respectively. The difference of the power by a factor of ~8 is contributed by different estimation methods. The different wavelength range (Hisaki EXCEED: 900–1,480 Å, Juno UVS: 700–1,190 Å, and 1,230–1,620 Å) and exclusion of the auroral emission at the geocoronal emission lines for Hisaki EXCEED contributed to the difference by a factor of 3 when the color ratio of Hisaki EXCEED (Tao et al., 2016; Tao et al., 2016; Tao et al., 2018) was 1.5–2 corresponding to the time interval of our interest. In addition, the longitudinal modulation correction causes the difference by a factor of 2, since Juno UVS refers to the maximum intensity of the total power, whereas Hisaki EXCEED refers to the average of the total power. Although the factor 1.3 discrepancy still remains, the aforementioned factors mostly explain the discrepancy between UVS and EXCEED. In the next section, we will discuss the temporal characteristic in more detail.
3 Results and Discussions
Figure 1 shows both transient and long-lasting (several-day) auroral enhancements. The transient event means that aurora suddenly becomes brighter with <10-hr duration (Kimura et al., 2015), and long-lasting event indicates that the total auroral power elevates for several days. Several-day enhancements have also been identified in past Hisaki observations (Kita et al., 2016) with solar wind compressions as a candidate to trigger an event of this category. Meanwhile, the transient event occurred around the arrival of a solar wind shock on DOY ~142 (Kimura et al., 2017; Nichols et al., 2017) as well as DOY 173. These transient brightenings could be triggered by solar wind disturbances.
Three significant events are named as Event 1 (DOY~142), Event 2 (DOY~150), and Event 3 (DOY~173), all shown in Figure 2. The panels on the upper show the total auroral power obtained from Hisaki (Figures 2a and 2b) and Juno/UVS (Figure 2c). The open circles are 10-min data corrected for longitudinal modulations, and the solid line indicates averaged total power over one Jovian rotation. The black dotted line identifies the 10-hr modulation, which indicates when the auroral oval faces Earthward (at the maximum of the curve). The green dashed lines indicate the HST observations (Nichols et al., 2017). The lower panels (Figure 2d, 2e, and 2f) are for solar wind dynamic pressure measurements. The dotted lines indicate the original solar wind data obtained from JADE observations, and solid lines show the solar wind data shifted all the way to Jupiter. It should be noted that this estimate assumes the propagation of the solar wind disturbances to Jupiter without considering the slowdown and diverting of the solar wind. Red vertical dashed lines indicate shock arrival time at the center of Jupiter (based on the list in McComas, Szalay, et al., 2017). The upper horizontal axis indicates the time from shock arrival. The Juno-Jupiter distance and Juno-Sun-Jupiter angle are shown in the bottom horizontal axes. The lag time between the expected shock arrival time at Jupiter location and aurora peak is 10–15 hr for Events 1–3. Obviously, the lag time between solar wind shock and aurora brightening cannot be explained solely by the ballistic propagation of the solar wind, because the lag time contains the information on the propagation time in the magnetosheath as well as the dynamical evolution of the magnetospheric plasma.
We also estimated the propagation speed of the solar wind disturbance in the magnetosheath, although the bow shock and magnetopause crossing time would be model dependent. We use the same model as Nichols et al. (2007). The solar wind velocity (vsw) decreases to 0.26vsw just downstream of the bow shock and then linearly decreases to 10 km/s at the magnetopause. The location of bow shock and magnetopause is calculated based on Huddleston et al. (1998). The several minute propagation time of the solar wind disturbances between the magnetosphere and the ionosphere (Cowley & Bunce, 2003b; Nichols et al., 2007) is negligible. Figure 3 summarizes the three major events using the same format as Figures 2a–2c. The horizontal axis indicates the time from the ballistic shock arrival at the center of Jupiter. The expected crossing times of the bow shock and the magnetopause (MP) are indicated by black arrows. The key parameters are summarized in Table 1.
| Pdyn (nPa) | Vsw (km/s)S | BS (RJ) | MP (RJ) | Arrival time at MP (model) | Aurora peak time (observation) | Lag time (hr) | |
|---|---|---|---|---|---|---|---|
| Event1 | 0.250 | 470.9 | 60.1 | 48.1 | 2.75 | 15.23 | 12.48 |
| Event2 | 0.032 | 457.5 | 100.2 | 75.5 | 6.79 | 12.76 (HST) | 5.97 |
| 14.75 (Hisaki) | 7.97 | ||||||
| Event3 | 0.740 | 413.2 | 45.8 | 37.9 | 1.61 | 12.49 (North) | 10.87 |
| 11.49 (South) | 9.87 |
- Note. The solar wind values (dynamic pressure (Pdyn) and velocity (Vsw)) and bow shock (BS)-magnetopause (MP) locations represent the postshock condition. The solar wind disturbance arrival time and aurora peak time are measured from the ballistic shock arrival at Jupiter.
For Event 1 (Figure 3a), when the shock arrived at the magnetopause (~3 hr after the ballistic shock arrival), the aurora did not respond. The aurora enhanced in the next rotation, about 10–12 hr after the shock arrival at the magnetopause. HST observed a dawn storm (Nichols et al., 2017, ~14 hr after the ballistic shock arrival). A similar trend can also be seen in Event 3 (Figure 3c); the aurora peak was located around 10 hr after the shock arrival at the magnetopause. Juno/UVS was off just around the shock arrival. We cannot rule out the possibility that the aurora becomes brighter one rotation before Event 3. It is plausible, though, that the aurora was quiet during the previous rotation as transient events observed by Hisaki EXCEED do not typically last more than two rotations (>20 hr) (Kimura et al., 2015). Unfortunately, there was no HST observation during that time, and Juno was too far from Jupiter to resolve the aurora. For Event 2, Hisaki detected the enhancement around DOY 150 (15 hr after the shock arrival), while HST showed that the dusk active region increased around DOY 149.9 (13 hr) (Nichols et al., 2017). The travel time between the bow shock and magnetopause was longer than in Events 1 and 3 since the magnetosphere was inflated owing to the weak dynamic pressure. The lag time for Event 2 is at most 6 hr.
It is still difficult to identify the brightening onset. The aurora light curve contains both brightening of the aurora and the apparent rotational modulation. If the aurora enhances at a longitude ranges which do not face the Earth, we cannot observe the enhancement. In the case of Event 1, we cannot rule out the possibility that the aurora became brighter about 3~8 hr earlier when Hisaki cannot see it. In the case of Event 2, the aurora onset might occur around the MP crossing time when Hisaki cannot see the aurora, which means that the propagation model (Nichols et al., 2007) fits well with the observation. Although there are some concerns, the lag time between the magnetopause and the brightening onset found here would be the key to understand how the Jovian magnetosphere, which has its own strong mechanism to drive its activity, responds to a solar wind shock.
It should also be noted that another shock on DOY 150.7 after Event 2 seems to trigger minor enhancement on DOY 151.2, which is smaller than the enhancement of Event 2 (Figures 4a and 4d). Here we discuss the possible causalities for these two events based on Kita et al. (2016) that the auroral total power shows a positive correlation with the duration of a quiescent interval of the solar wind. The quiet time for Event 2 is assumed to be ~3.5 days (DOY 146–149.4), while the quiet time of the minor enhancement on DOY 151.2 is ~1.3 days (DOY149.4–150.7). According to quiet time-aurora total power relationship in Kita et al. (2016), it is plausible to cause 100-GW enhancement with 3.5 days quiet time, whereas the enhancement with 1.3 days quiet time is less than 100 GW. However, the quiet time of Event 1 is only 2–3.5 days, but the total power increases more than 1,000 GW (Figures 4b and 4e). As Nichols et al. (2017) pointed out the enhancement of volcanic activity before Event 1 (de Kleer & de Pater, 2017; Tsuchiya et al., 2019), one hypothesis is that the increased magnetospheric plasma content might cause large aurora enhancement. The positive relation between the quiet time and the aurora total power suggests that the aurora total power depends on the magnetospheric plasma content (Kita et al., 2016). The volcanic enhancement before Event 1 might mean that a large amount of plasma was already supplied to the magnetosphere and the aurora could be brighter without a long quiet time.
Next, we focus on the threshold of the solar wind shock to trigger the auroral enhancement. As shown in McComas, Szalay, et al. (2017), Juno/JADE detected several solar wind shocks during the Juno approach phase. From the criteria used in Kita et al. (2016), it is possible to find aurora enhancement if the event has a long quiescent interval and dynamic pressure is larger than 0.11 nPa. Figures 4a and 4d are Event 2 (DOY 150), which is the same plot as Figures 2b and 2e. Event 2 seemed to be triggered by the shock around DOY 149.5 when the dynamic pressure elevated to around 0.07 nPa. The next solar wind shock arrived on DOY 150.7, which was later than the aurora enhancement. On the other hand, the dynamic pressure around DOY 168 increased to ~0.14 nPa (Figures 4c and 4f), which was larger than that of Event 2. However, DOY 168 event seems to be a minor UV enhancement in spite of long quiescent interval (~15 days). From UVS observations, minor enhancement occurred around DOY 168, which was almost buried in the increasing trend of the northern aurora. Hisaki EXCEED also observed the small enhancement around that time. Moreover, the south aurora did not respond to the solar wind disturbance. Therefore, DOY 168 event seems not to be a significant variation in auroral power. This result may indicate that the variations of other parameter have more correlation with the total power of aurora. Not only the dynamic pressure but also other parameters (magnetic field strength, reconnection voltage, etc.) increase during the solar wind compressions. Although the number of events is not enough to conclude the exact parameter correlated with the aurora brightening, these results bring a question on the aurora response to the solar wind shock.
It is also important to compare these results with numerical simulation, though the realistic input of the upstream solar wind condition is necessary for direct comparison with the Hisaki/Juno results. Several authors reported the response of Jovian aurora to the transient solar wind variation. Three-dimensional Global MHD simulation done by Chané et al. (2017) shows that the aurora current gradually increases after 2.5 hr and reaches a new equilibrium after 60 hr, where the start time corresponds to the moment when the high-density region arrives at the bowshock. The solar wind input is relatively similar to Event 2 in that the dynamic pressure elevated more than 60 hr. The discrepancies between the observation and the model are that the observed aurora was still quiet after 2.5 hr and became brighter at most 17 hr after the shock arrival at the bowshock. Cowley et al. (2007) and Yates et al. (2014) reported that the magnetosphere-ionosphere coupling current increased during the transient solar wind compression. Yates et al. (2014) used the time evolution of magnetopause location represented by a Gaussian; the magnetopause location becomes minimum after 1.5 hr and then returns to the expanded position after 1.5 hr. That is, at the moment when the magnetopause location becomes minimum corresponds to the MP crossing time shown in Figure 3. The lag time between the compression and the peak of aurora current is not shown in Yates et al. (2014), but it would not be greater than 1.5 hr. Therefore, the onset time of Event 2 does not contradict with Yates et al. (2014). On the other hand, the transient type brightenings (Events 1 and 3) have a lag time of ~10 hr, which cannot be explained by these models. Again, the continuous solar wind parameters are available during this period, so the numerical simulations with the realistic solar wind input can directly compared with Hisaki EXCEED and Juno UVS data.
4 Conclusions
During Juno's approach phase to Jupiter, we found that the aurora triggering process is not simple; the lag time of the transient events cannot be explained by the propagation model of the solar wind disturbances in the magnetosheath, whereas that of the several days brightening event does not contradict to the propagation model. In addition to that, the size of the dynamic pressure enhancement does not affect the variation in the aurora total power. Solar wind compression also changes the current system in the middle magnetosphere and affects the electric field in the inner magnetosphere, which appears as the variation in the dawn-dusk asymmetry of Io plasma torus (Murakami et al., 2016). During the Juno approaching phase, the dawn-dusk asymmetry varied with the three major shocks, which also provide significant information on the magnetospheric response to a solar wind shock. Although three events are not enough to determine the solar wind control of the Jovian magnetosphere, it is also important to compare with theories and numerical simulations. Notably, owing to coincident Juno measurements of the upstream solar wind and Hisaki observations of the auroral regions, the lag time between solar wind shock and aurora enhancement found from this study is unambiguous and can be used to test theoretical and numerical descriptions of Jupiter's auroral response to solar wind driving.
Acknowledgments
The data of Hisaki satellite are archived in the Data Archives and Transmission System (DARTS) of ISAS/JAXA (http://darts.isas.jaxa.jp). The Juno Solar Wind data used in this study may be found in the supporting information section of Wilson et al. (2018). Juno UVS data are archived in NASA's Planetary Data System (http://pds-atmospheres.nmsu.edu/data_and_services/atmospheres_data/JUNO/juno.html). The authors acknowledge the support of ISSI, as this study was discussed within ISSI International Team “The influence of Io on Jupiter's magnetosphere”. This work was supported by JSPS KAKENHI grants JP26287122, 19H01948 and 17H02965. H. K. was supported by Grant-in-Aid for JSPS Research Fellow. F. T. was supported by JSPS KAKENHI grant 26400476. T. K. was supported by JSPS KAKENHI grant K16K178120.
